All HBV- and HCV- notifications in the years 1990 to the end of 2004 were identified. In the HBV cohort all the notifications for acute hepatitis B and those with HCV co-infection were excluded. In addition, all individuals who were reported to have been infected in adulthood were excluded. All HBV notifications were matched with the HCV notifications from 1990 to 2004 to identify co-infected individuals. After exclusion of all notifications of acute HBV infection, the remaining ones were included in the HBV–HCV co-infection cohort. The study population consisted of two cohorts; the HBV-cohort consisted of 9,646 subjects with chronic HBV infection and the HCV-HBV co-infection cohort consisted of 1,697 patients.
3.2.2 Paper II
All HBV- and HCV-notifications from years 1990 to 2003 were identified. All notifications of acute hepatitis B were excluded. The study population constituted of three cohorts; the HCV-cohort consisted of 34,235 individuals reported for HCV-infection (no HBV infection), the HBV-cohort of 9,517 individuals reported with chronic HBV infection (no HCV infection), and the HCV-HBV co-infection cohort consisted of 1,601 individuals with both HCV and chronic HBV-infection.
3.2.3 Paper III
All HCV notifications to the PHAS years 1990-2011 were identified. Identification of all childbearing women with HCV infection and their infants was obtained by linking information on all women notified with HCV infection to MBR by using their PIN. The medical birth register then matched the infants to five controls, i.e. infants of mothers without HCV infection. The matching criteria were birth year, gender, county of origin and age of the mother. Because of the number of matching criteria, the total (mean) numbers of controls per case were only 4.5. The study population constituted of 19,072 infants of 9,599 mothers reported with HCV during 1990-2011. They were compared with 86,164 infants of 83,986 mothers without HCV infection.
3.2.4 Paper IV
As in paper III, the study population consisted of all children of women notified with HCV infection in the National Surveillance Database at the PHAS years 1990 to 2011. These children were matched with five controls, i.e. infants of mothers without HCV infection. The matching criteria were birth year, gender, and county of origin and age of the mother. Because of the number of matching criteria, the total (mean) numbers of controls per child were only 4.5. The study population constituted of 19,097 offspring of 9,599 mothers with HCV infection compared with 86,192 children of 83,986 mothers without HCV infection.
3.3 ANALYSIS 3.3.1 Paper I
The study period was 1990-2004. The PINs of individuals in the HBV and HBV-HCV co- infection cohorts were used to link to the Cancer register. The National Tax Board added information on dates of emigration, immigration, deaths, and country of birth.
To avoid overestimation of the risk due to surveillance bias, the time of observation commenced three months after the date of HBV notification for each subject in the HBV
study population. For the HCV–HBV co-infection cohort, the observation time commenced three months after the date of the second of the two notifications. The observation time terminated either at death, the first date of HCC diagnosis reported to the Cancer Registry, or 31 December 2004, whichever came first.
Chronic HBV infection in Sweden is mostly found in immigrants from high endemic areas (259, 260, 262). 90% of all patients that are infected with HBV during the perinatal period develop chronic infection. We therefore stratified the cohort according to age; less than 30 years, 30 to 39 years, 40 to 49 years , 50 to 59 years, and 60 years and older. Earlier epidemiological studies have demonstrated that the HCV-HBV co-infection cohort is similar to the HCV cohort in Sweden concerning transmission route. We therefore estimated time of infection using a model developed in earlier comparable studies on the HCV cohort (135, 284). Time from infection was assessed for all routes of HCV infection based on available epidemiological data in Sweden (286-288). For transmission route notified as IDU, unknown or sexual transmission the year of infection of persons born before 1930 were considered to be in 1965; the age of infection in persons born in 1930 were considered to be at the age 35 years, then falling linearly so when born in 1955 or later they were considered infected at 20 years of age. We used the age at the time of notification for patients younger than 20 years. We approximated the year of infection to be 1980 for persons with transfusion-associated HCV infection before blood donor screening was introduced in 1991. Date of notification was used as the date of infection for persons with nosocomial and occupational route of transmission. We then made three strata depending on the time with infection: infection for less than 20 years, infection between 20 and 30 years, and infection for more than 30 years. We then assessed the risk of hepatocellular cancer (HCC) in the study cohorts by comparing with the general population. The expected number of HCCs was calculated on the basis of age, sex, and calendar year-specific cancer incidence rates from the Cancer Register, and the
standardized incidence ratios (SIR) of HCC by comparing the expected incidence with the observed HCC incidence. Ninety-five percent exact confidence intervals (CIs) were calculated assuming a Poisson distribution of the number of observed cancers.
3.3.2 Paper II
The study period was 1990-2003. The PINs of individuals in the HCV and HBV cohorts were used to link to the DR. The National Tax Board added information on dates of emigration, immigration, deaths, and country of birth.
For each subject, the observation time started six months after the HBV or HCV notification (first notification if co-infected) and ended at death, emigration, or the end of study, whichever occurred first. To avoid overestimation of the risk due to surveillance bias from HBV and/or HCV infections diagnosed as a result of the disease that led to death, all individuals who died less than 6 months after the hepatitis notification were excluded.
ICD codes according to the main ICD-chapters were used to categorize cause of death in main groups. ICD codes of special interest (liver related, drug and alcohol related, external reason, HIV, NHL, MM and other malignancies) were then further analyzed. The standardized mortality ratio (SMR) was calculated by comparing the mortality in the study population with the mortality in the general population. The observed number of deaths was divided by the expected number of deaths. The mortality rates for the general population were obtained from the DR. For the calculation of the expected number of deaths, the sex and age- specific mortality rates in the calendar year 1999 were used.
3.3.3 Paper III
The study period was 1973-2011. We used relevant ICD codes from ICD-10 (1997-2011), ICD-9 (1987-1996) and ICD-8 (1973-1986) for the studied outcomes. Information about pre- pregnancy diagnoses of the mothers, such as diabetes, hypertension, alcohol and drug use,
information on the number of years of formal education completed at the time of birth of the child, categorized as: less than 9 years, 9 to 12 years, more than 12 years. Country of birth was categorized into: Sweden, other Nordic country, non-Nordic European country and non- European country. Cigarette smoking at any visit to the maternity care was reported as daily smoking. Women were categorized by whether they were living with the child’s father, not living with the father, or other family situations. Women were categorized as lean if body mass index (BMI) at first antenatal visit was 11.0- 19.9 kg/m2, normal weight if BMI was 20.0- 24.9 kg/m2 , overweight if BMI was 25-29.9 kg/m2 and obese if BMI was 30.0- 60.0kg/m2. Parity was divided into 1, 2 or 3+ children. Data on complications of the pregnancy (preeclampsia, gestational diabetes, gestational hypertension), were obtained from the MBR.
Information on outcomes regarding the infant was obtained from the MBR such as Apgar score, birth weight and gestational age. Apgar score at 5 minutes after birth was used, categorized into 0-6 and 7-10 (used in earlier comparable studies (289)). Gestational age at birth was categorized into: very preterm (<32 weeks), moderately preterm (32-36 weeks), and term (≥37 weeks). The infants were then categorized if they were small for gestational age (SGA). SGA was defined as birth-weight less than two standard deviations (SD) below the mean for gestational age based on Swedish reference curve of estimated fetal growth (290). Low birth weight was defined as <2500 grams. Gestational age at birth was categorized into: very preterm (<32 weeks), moderately preterm (32-36 weeks), and term (≥37 weeks). For outcome of interest, relevant ICD codes from ICD-10 (1997-2011), ICD-9 (1987-1996) and ICD-8 (1973-1986) was used. Information on malformation was obtained from the MBR and from the PAR. Further information on complications at birth, such as cephalohematoma, neonatal seizure and intraventricular haemorrhage was obtained from the MBR. Information over intrauterine death and postpartum death was obtained from the MBR, which retrieved
stillborn after 28 weeks of gestation. In 2008 NBWH changed their definition to stillborn after 22 weeks of gestation. Neonatal death was categorized into early neonatal death (within 6 days) and late neonatal death (7-27 days).
To investigate the relationship between HCV and the binary outcomes of pregnancy and births, adjusted odds ratio (aOR) were calculated through multivariate logistic regression. To estimate the association between HCV, gestational age and the Apgar score, adjusted odds ratios were calculated with linear regression.
For the models concerning gestational diabetes, hypertension, preeclampsia, caesarean section, still birth, early and late neonatal death, malformation and SGA, the following covariates were adjusted for: parity, alcohol use of the mother, drug use of the mother, diabetes and hypertension diagnoses prior to pregnancy, smoking, BMI of the mother, educational level of the mother and family situation. The covariate was included in the model if the interaction effect was more than 10%. P values <0.05 were considered to be statistically significant.
3.3.4 Paper IV
The study period was 1973-2011. Identification of all childbearing women with HCV infection and their infants was obtained by linking information on all the women reported to PHAS to the MBR, using their PIN. For this study we used the MBR and the DR. Information on outcomes regarding the infant was obtained from the MBR. Gestational age at birth was categorized into: very preterm (<32 weeks), moderately preterm (32-36 weeks), and term (≥37 weeks). The infants were then categorized if they were SGA. SGA was defined as birth-weight less than two standard deviations (SD) below the mean for gestational age based on Swedish reference curve of estimated fetal growth (290). Statistics Sweden added information on social economical support of the mother and her family at the time of birth.
As in paper III, the relevant ICD codes from ICD-10 (1997-2011), ICD-9 (1987-1996) and ICD-8 (1973-1986) were used for outcome of interest. ICD codes which constituted the main ICD-chapters were used to categorize cause of death. ICD codes of special interest (liver related, drug and alcohol related, accidents, Sudden Infant Death Syndrome (SIDS)) were then further analyzed. Information on date and cause of mortality of the children was obtained from the DR. The baseline characteristics between children of HCV-affected mothers (cases) and children of non-affected mothers (controls) were compared by reporting absolute and relative frequencies. Differences between the two groups were tested for with the chi-square test.
Survival time was analyzed. The end of follow-up, December 31, 2011, was considered an independent censoring event. Crude survival curves in the two groups were calculated with the Kaplan-Meier method. Crude and adjusted mortality HRs were estimated with proportional-hazard regression models. The crude model included the binary indicator for case (0 = control, 1 = case) as the only covariate. The adjusted models also included indicators for low gestational age, low birth weight, SGA, and smoking during pregnancy. P- values less than 0.05 were considered statistically significant.
3.4 ETHICAL APPROVALS
For all papers the personal identifiers were removed before the dataset was used for analysis. All studies were approved by The Regional Ethical Review Board in Stockholm according to the guidelines of the Helsinki Declaration.
4 RESULTS
4.1.1 Paper IA total of 15,318 HBV infected patients were reported to SMI between the years 1990 and 2004. After exclusion of all acute HBV and all HCV co-infected individuals, the HBV cohort consisted of 9,464 subjects with chronic HBV-infection. The HBV study population contributed to overall 66,768 person-years of observation overall. 85% were born in 1950 or later and males accounted for 53%. A total of 76 patients were reported to the Cancer register with cancer originated in the liver. Of them, 31 patients were excluded because the cancer diagnosis was made before or within three months after the report of the HBV diagnosis. This left 45 patients for further risk analysis. The mean time between the HBV notification and the cancer diagnosis was 4.8 years. The SIRs are presented in table 1. The lifetime risk of developing HCC was estimated to be 10% (95% CI: 6-12)
Table 1. SIRs for HCC (n=45) in HBV infected patients
Years with HBV infection Expected Observed SIR 95% CI
0-29 0.03 1 33 0.9-189 30-39 0.08 3 38 7-105 40-49 0.19 9 47 21-89 50-59 0.35 19 54 33-85 >60 0.65 13 20 11-34 Total 1.3 45 35 25-46
There were 3,238 patients reported with co-infection between 1990-2004. After excluding those with acute hepatitis B (n=1,850), the co-infection cohort consisted of 1,697 patients in the risk analysis. Observation time was 11,392 person-years. A total of 67% were born after 1950 and males accounted for 78%. In the co-infected cohort, 12 patients were diagnosed with HCC and reported to the Cancer register. Two of them were excluded because they were diagnosed with HCC within three months from diagnosis of HCV. This left 10 patients for further risk analysis. Median age of these patients was 58 years. The standardised incidence
Table 2. SIRs for HCC (n=10) in HBV and HCV co-infected patients Years with co-infection Expected Observed SIR 95% CI
0-19 0.025 0 0 0-150
20-30 0.118 4 34 9-87
>30 0.066 6 91 33-198
Total 0.209 10 48 23-88
4.1.2 Paper II
The mean observation times per subject in the HBV cohort was 6.4 years, contributing to totally 60,697 person years. Mean age at death was 56 years, 68% were males. The most common route of transmission was neonatal transmission (80%). After excluding all individuals that died within six months from HBV notification there were 425 (4.5%) deaths. The all-cause mortality was significantly increased with SMR 2.3 in the HBV cohort. The most frequently underlying cause of death in the HBV cohort was neoplasm. The cause of death for all cohorts is presented in table 3.
The mean observation time of the HCV cohort was 6.3 years contributing to 214,602 person years. The mean age at death was 50 years and 77% were men. The most common route of transmission was IDU (57%) or unknown (32%). After excluding all deaths within six months from HCV notification (n=744) there were 4,651 deaths. The results are presented in table 3. The all-cause mortality was significantly increased with SMR 5.8 in the HCV cohort. The most frequent underlying cause of death in this cohort was external causes (e.g. injuries, intoxication, and suicide) which accounted for 29% of the deaths.
The mean time of observation in the HCV-HBV co-infection cohort was 7.9 years, in total 12,667 person years. The mean age at death was 44 years and 85% were male. The most common route of transmission was IDU (54%) or unknown (25%). There were 209 (13%) deaths in the co-infected cohort after excluding all deaths reported within six months from first HBV/HCV notification (n=21). The most frequently reported cause of death was the
same as for the HCV cohort or external causes (e.g. injuries, intoxication, and suicide), 34 %. The all-cause mortality was also significantly increased in this cohort with SMR 8.5.
In the HCV and the co-infected cohort, there was an excess risk for death from causes related to IDU, e.g. HIV, psychiatric diagnoses (98% drug related) and external reasons, compared with the general population. However, the relative risk of liver-related mortality was highly increased in all three cohorts.
Table 3. Cause of death in the HBV, HCV and HBV-HCV cohorts. The risk is expressed as SMR, the observed deaths/expected deaths
HBV (311 deaths) HCV (3,970 deaths) HBV-HCV (188 deaths)
Diagnosis SMR* 95% CI SMR* 95% CI SMR* 95% cI All Cause 2.3 2.0-2.6 5.8 5.6-6.0 8.5 7.3-9.8 Infection 13.7 8.7-20.6 28.7 25.2-32.5 44.8 25.1-74.0 Neoplasm 2.8 2.3-3.3 2.8 2.6-3.1 3.9 2.5-5.9 Blood/immune 6.3 0.8-22.6 30.6 22.6-40.6 19.6 0.5-109.2 Endocrine 1.8 0.7-3.8 5.6 4.5-6.9 6.6 1.8-17.1 Psychiatric 3.1 1.8-5.0 15.0 13.7-16.5 26.0 18.0-36.3 Circulatory 1.4 1.1-1.8 2.6 2.4-2.8 5.1 3.4-7.3 Digestive tract 5.8 3.8-8.5 15.3 13.8-16.9 16.5 9.0-27.6 External reasons 1.7 1.2-2.4 12.4 11.7-13.1 11.4 8.8-14.6 Subgroups Viral hepatitis 78.9 46.8-124.7 133.0 114.3-153.9 168.6 84.2-301.7 Liver cancer 31.2 21.9-43.2 34.9 30.1-40.2 65.2 33.7-113.9 Liver disease 10.7 6.8-15.9 25.1 22.3-28.0 24.4 13.0-41.8
All liver related 21.7 17.1-27.0 35.5 32.9-38.3 46.2 31.5-62,3
HIV 11.4 2.4-33.2 41.2 31.4-53.2 23.7 2.9-85.5
Alcohol,drug related 3.8 2.1-6.4 20.7 18.9-22.7 27.6 19.6-39-6 *SMR= observed/expected, the expected number of deaths were calculated using age and sex specific mortality rates in the general population
All cause SMR related to age showed an excess mortality in all ages, but in the HCV and HBV-HCV cohorts the great excess mortality was at age 15-35, then slowly declining. The SMR by age showed that the excess mortality from liver related deaths increased with age. The maximum liver related SMR was 26 at age 60-69 in the HBV and 42 at age over 70 years in the HCV cohort.
The liver cancer mortality had the greatest excess risk with SMR 31.2 (HBV), 34.9 (HCV) and 65.2 (HBV-HCV). In the HCV cohort risk of death from cancer in lip/mouth, oesophagus, pancreas, larynx, lung, cervix uteri and kidney was also significantly increased.